Implications of geitonogamous pollination for floral longevity in Iris gracilipes

1. We analysed the longevity of individual flowers with respect to the effects of the floral display size of a plant (number of flowers open at one time) and spatial arrangement of flowers in Iris gracilipes (A. Gray). We found that the actual floral longevity of this species (about 2 days) was much shorter than that predicted by the model of Schoen and Ashman (about 7-10 days). 2. A large display size and an overlap of flowering periods within a shoot facilitated successive pollinator probes within a plant, though the frequency of pollinator visits per flower was independent of display size: Consequently, the cost of geitonogamous selfing increased with an increase in display size and an overlap of flowering periods within a shoot. The short floral longevity of this species would be effective in such situations because shorter floral longevity leads to a smaller display size and a smaller overlap of flowering periods within a shoot. 3. A small display size and small overlap of flowering periods within a shoot could also be realized by the strategy that staggers the flowering period of each flower without floral longevity being shortened (extended blooming). However, a mutant plant with an extended blooming schedule would be unlikely to invade the population because of the stabilizing selection pressure on the flowering date. This may be why such a short floral longevity, rather than extended blooming, has evolved in this species.